SlideShare une entreprise Scribd logo
1  sur  25
Télécharger pour lire hors ligne
黄浴 | 首席科学家
仿真测试中的数据闭环
如何打造自动驾驶
2
背景介绍
DARPA 越野挑战赛 (2004-2005)
Stanford冠军
---谷歌团队 GoogleX
DARPA 城市挑战赛 (2007)
CMU冠军
---Uber团队 ATG
自动驾驶的“上”和“下”半场
L2-L3-L4-L5:渐进升级/一步到位/降维(量产落地)
如何实现自动驾驶的技术工程化落地,包括开发标准化和平台化、量产规模化
和落地商业化(成本、车规和OTA)。
“下半场”意味着要解决罕见的”长尾”场景,同时构建数据闭
环的持续高效研发框架
“上半场”大多是解决常见的问题
全栈技术:感知、定位、预测、规划决策和控制在典型场景(即高速、街道和
停车场等)的解决算法和执行方案(线控底盘)。
计算平台(AI芯片及其SOC): 比如英伟达的Xavier和Orin
传感器技术: HDR摄像头、固态激光雷达和4D毫米波雷达等。
两种开发路线走向合作。
自动驾驶的算法框架
模块化:
局部端到端:
端到端:
安全
安全是一个系统级概念,目的是将系统组件故障引起的事故风险降至最低
AI安全是一个各种机器学习漏洞引起的新问题,需要解决:=》数据闭环;
功能安全标准(ISO26262):
确定安全需求,定义安全要求,并
最终验证相应的设计;
SOTIF (预期功能安全)
规避性能局限或滥用带来潜在危害
的不合理风险;
安全模型:
Mobileye提出的责任敏感安全模型(RSS)
Nvidia’s 安全力场模型 (SFF)
存在的问题:
① 极端案例
② 对抗性攻击
③ 可解释性
④ 不确定性
⑤ 验证和确认(V&V)
数据闭环
AI的最挑战应用之一,自动驾驶,是一个长尾效应的典型。
极端情况(corner case)往往是要搜集的数据,发现这些有价值的数据,标注后放入训练集,同时放入测试集或者仿真场景库;模型得到迭代升
级后,再交付(模型压缩蒸馏)到车端进入新的循环,即数据闭环。
数据筛选
研发阶段和量产阶段的不同;
车端:
ü 异常驾驶;
ü 影子模式;
ü 相似场景或者事件搜索;
服务器端:
ü 主动学习;
• 不确定性估计;
ü 场景挖掘:
• 安全-紧要。
数据筛选
主动学习
ü 其⽬标是找到有效的⽅法,从未标记的数据池中选择要标注的数
据,最⼤限度地提⾼准确性。
ü 通常是⼀个迭代过程,每次迭代⽤⼀些筛选准则(如不确定性估
计)从未标注池中选择⼀组数据进⾏标注。
标注所
选样本
基于标注数据
训练模型
添加新标注样
本到训练数据
筛选函数从
⽆标注数据
中选择样本
ü 不确定性估计
✘ 不确定性指知识有限的状态,无法准确
描述现有状态、未来结果或不止一种可
能的结果;
✘ 包括认知/模型(epistemic)和任意/数据
(aleatoric)不确定性;
Ø Monte Carlo Dropout;
Ø Bayesian Learning;
Ø Ensemble Learning。
数据筛选
关键场景识别 (CSI) :查找触发条件、安全-紧要
操作情况或将导致伤害的两者组合方法。
✓ 1 设计空间探索(案例研究 )
! 无轨迹:基于搜索的采样、优化和测试等;
! 有轨迹:路径规划、强化学习和优化等;
✓ 2 归纳推理(基于测试场景或者事故场景)
✘ 聚类、文本分析、重建和仿真等。
✓ 3 演绎推理(基于各种知识源寻找预碰撞 )
✘ 一组预定义的假设下,系统地考虑所有可能性。
✓ 4 基于感知理解找到关键景象
✘ 景象表征
Ø 基于描述
✘ 关键景象生成或者探索
Ø 直接转换
Ø 基于优化
Ø GAN
Ø 场景描述
Corner case – Edge case - OOD - Anomalies
数据标注
自动标注是一个提高数据闭环效率的重要模块
ü 离线, 非实时,高性能;
ü 服务器/云端(而不是车端);
ü 可以不局限于机器学习或者深度学习模型;
ü 数据输入可以不是单帧(单时间步);
标注工作耗时并且昂贵(某些真值不是人工可以获得)
半自动标注
ü “人在环”的交互
ü 依赖比人工操作更可靠的算法,比如3-D重建
高清地图的制作是其特殊的一部分
仿真数据自带标注
数据标注
BEV模型及其类型
ü BEV是鸟瞰图(Bird’s Eye View)的简称,也被称为上帝视⾓,是⼀种⽤
于描述感知世界的视⾓或坐标系(3D);在自动驾驶的系统中,规划和
控制一般在BEV空间进行。
ü 基于BEV的视觉感知系统即代表⼀种端到端的、由神经⽹络将摄像头透视
投影的视觉信息从图像空间转换到BEV空间的环境理解框架。
ü BEV的视觉感知中关键的部件是视角变换。根据变换的方式,可以将BEV
方法粗分为两个大类,即基于几何和基于网络的方法。
“Delving into the Devils of Bird’s-eye-view Perception”
“Vision-Centric BEV Perception: A Survey”
ü 虽然基于硬平地假设,但基于单应变换的方法具有良好的可解
释性,其中IPM在感知中的图像投影或特征投影中发挥作用;
ü 基于深度的方法通常基于明确的3D表征,即量化体素或点云;
✘ 基于点的方法存在模型复杂、性能低等问题;
✘ 基于体素的方法由于计算效率和灵活性很受欢迎。
ü 由于缺乏深度信息、遮挡等原因,基于MLP的方法不方便;
ü 具有稀疏或密集查询的Transformer,凭借强大的相关建模和数
据依赖特性获得了最佳性能,但效率仍然是一个问题。
ü 数据标注:3-D 特性
ü 感知分辨率:感知距离目标大小
ü 网络架构:主干 (RegNet) /颈部 (FPN)
ü 多头结构: 多头
ü 共享主干?
ü 多任务学习(MTL):
ü 附加任务设计, 多阶段学习
ü 扩展: 感知->预测->规划 & 决策
ü 优化/压缩: 蒸馏
BEV方法比较
BEV方法关注点
数据标注
BEV的真值GT
NuScenes
图像
IMU
里程计
GPS
线下
非实时
神经网络
模型
分割
深度
流
静态背景及自车轨迹
活动目标及其运动参数
Tesla
路面高度
数据标注
静态目标标注
动态目标标注
激光雷达目标跟踪
ü 激光雷达Multi-view fusion (MVF) 目标检测增强方法:MVF++
激光雷达目标检测
Google WayMo
模拟仿真
Cruise. AI
Tesla
传感器仿真
图像/视频的渲染
传感器模型(摄像头,激光雷达)
交通流仿真
SUMO
AI
道路环境仿真
高清地图
数字孪生
道路参与者仿真
车辆,行人,自行车/摩托车…
运动学/动力学模型
实-虚
神经渲染(包括NeRF)
虚-实
风格转换(GAN)
仿真降低自动驾驶测试成本
模拟仿真
神经渲染-NeRF基础
ü 神经渲染是将经典(基于物理学的)计算机图形学与深度学习相
结合。其目标是以一种可控的方式生成照片般真实的图像;
ü 一种范式是神经网络被训练为从一些2-D输入信号直接映射到输
出图像,即把神经网络训练为一个渲染引擎;
ü 另一种范式是一个神经网络代表一个特定场景的形状或外观,并
且代表用一个传统的图形学 “引擎 ”进行渲染;
ü 神经辐射场(NeRF)是一种基于坐标(3-D位置和2-D观察方向)
的神经网络采用体渲染技术输出对应的颜色和体密度方法;
ü 必要的两个优化策略:
ü 位置编码:映射到高维空间,适用于细节;
ü 分级体采样:同时优化粗网络和细网络。
ü 训练和渲染速度都很慢;
ü 只能表示静态场景;
ü 固定照明;
ü 训练的NeRF表征途不会推广到其他场景/目标。
基础NeRF的不足:
模拟仿真
ü 分解场景即其MLP表征;
ü 在稀疏3D网格做缓存,进行体素裁剪和内插,输入以球谐波系
数取代观察方向;
ü 有效地预计算沿光线的部分或全部体积分;
ü 采用网格、稀疏网格、树和哈希等数据结构;
NeRF加速方法
NeRF重打光方法
ü 将场景表征为体密度场、表面法线和BRDF等,以此模拟光透
射的过程,支持任意照明条件下渲染场景。
动态场景的NeRF方法
ü 一是时变表征方法,可以处理一般目标和场景;
ü 二是控制变形状态的技术,使用特定类别的运动模型。
✘ 隐式地在变形上调节辐射场;
✘ 显式地用单独变形MLP去扭曲(warp)空间,回归变形空间
到静态规范空间的偏移。
模拟仿真
NeRF的泛化
ü 一类工作类似于基于图像渲染(IBR);
ü 另一类方法旨在学习场景的整体、全局表征,而不是依赖图像
或其他离散空间数据结构;
ü 还有一个类似方法,把一个备有神经渲染器的3D场景表征嵌入
到生成对抗网络(GAN)中。
NeRF优化摄像头姿态估计
ü Inverting Neural Radiance Fields for Pose Estimation(iNeRF);
ü Bundle-Adjusting Neural Radiance Fields (BARF);
NeRF渲染质量改进方法
ü NeRF-W 引入appearance embedding的优化方法,提出两个浅层
编码(表观嵌入和瞬态嵌入)来建模静止+暂态(不确定)效果;
ü 引入ISP来进一步优化表观特征,即ADOP;
ü Mip-NeRF 渲染抗锯齿conical frustums而不是NeRF的射线 ;
ü Mip-NeRF360 继续优化图像采样和混叠,合成无边界感的场景;
ü Raw-NeRF在原图像进行训练,保持场景HDR特性;
ü Block-NeRF:将大场景分解为小场景,每个单独训练NeRF。
模拟仿真
EPE(Enhancing Photorealistic Enhancement)
ü 卷积网络利用传统渲染流水线生成中间表征G-buffers,提供景深、
形状、光照、透明度、材质等特征信息;
ü 通过对抗目标训练,在多个感知层提供监督信号,用GAN的鉴别器
评估增强图像的真实性;
ü 在训练过程中对图像块进行采样,以消除图像伪影;
ü 改进1: 普通卷积取代strided convolution;
ü 改进2: 渲染-觉察非规范化(rendering-aware denormalization, RAD)
模块,取代BN层。
G-buffer encoder
Image enhancement network
Perceptual discriminator
LPIPS = Learned Perceptual Image Patch Similarity
Intel
模拟仿真
GeoSim是一种几何-觉察图像合成过程,从其他场景提取动态
目标以新姿态渲染,增强现有图像,合成新自动驾驶场景。
ü 构建一个具有真实几何形状和传感器数据
外观的多样化3D目标库。
ü 在仿真过程中,合成方法完成几何-觉察的
模拟工作:
✘ 1) 在给定场景放置合理真实的目标,
✘ 2) 从资产库中得到动态目标新视图,
✘ 3) 合成和混合渲染该图像。
ü 生成的合成图像具有真实感、交通-觉察和
几何一致性,允许扩展到复杂用例。
ü 跨多个摄像头传感器进行远程逼真视频模
拟,数据生成结果做下游任务数据增强。
“GeoSim: Realistic Video Simulation via Geometry-Aware Composition for Self-Driving”
SN-PatchGAN
基于场景的自动驾驶测试
一个场景(scenario)是自动驾驶组件及其驾驶环境在一段时间的动态描述;
从真实世界提取感兴趣的场景并生成失败案例对测试非常重要;
基于危害(Hazard)的测试方法,选择那些“智慧里程(smart miles)”,即体现自动驾驶车辆失败的安全-紧要场景 ;
对抗攻击:自动生成安全-紧要的场景。
测试场景
道路 环境(天气) 动态元素
脚本化交通 非脚本化交通
基于场景的自动驾驶测试
安全紧要场景生成
直接采样
基于知识的生成
对抗生成
数据驱动生成
静态场景 知识指导的学习
预定义规则
动态场景
密度估计
要生成安全-紧要场景,最重要的因素是场景的风险,通过交互双方之间的距离来描述,距离小意味着碰撞风险高。
ü 数据驱动方法仅用收集的数据直接或通过生成模型进行采样;
✘ 简单采样:数据重放、聚类和随机扰动;
✘ 密度估计:贝叶斯网络、深度学习模型和深度生成模型。
ü 对抗方法使用来自仿真部署的自动车辆反馈;
✘ 控制场景的初始条件;
✘ 通过策略模型序列地控制动态目标;
ü 基于知识方法主要利用来自外部知识的信息作为生成的约束或指导:
✘ 基于规则:手工设计场景的结构和参数 ;
✘ 基于学习:假设可以从专家获得特定领域知识。
场景生成存在的五个挑战
准确性
效率 多样性
可迁移性 可控性
数据驱动
对抗
基于知识
基于场景的自动驾驶测试
ü AdvSim是一种对抗性框架,为激光雷达自动驾驶系统生成安全关键场景;
ü 给定初始交通场景,AdvSim以物理上合理的方式修改参与者的轨迹,并更新激光雷达传感器数据;
ü 通过从传感器数据进行直接模拟,获得对全自主驾驶安全-紧要的对抗场景。
“AdvSim: Generating Safety-Critical Scenarios for Self-Driving Vehicles”
基于场景的自动驾驶测试
“AdvDO: Realistic Adversarial Attacks for Trajectory Prediction”
AdvDO(Adversarial Dynamic Optimization)是一个对抗性攻击框架,用设计的可微分动态模型生成真实对抗性轨迹。
1. 动态参数估计阶段:目标是从数据集采样的轨迹重建
真实的密集轨迹并估计正确的动力学参数给对抗轨迹生
成阶段。其中利用可微分动力学模型优化控制动作。
2. 对抗轨迹生成阶段:目标是生成一条对抗性轨迹,在
给定的限制条件下误导未来的轨迹预测。为实现这一目
标,设计带多个正则化损失约束的对抗损失函数 。
总结
①目前自动驾驶的发展面临挑战;
②数据闭环平台的实现功能,体现开发者的能力;
③获取/挑选有价值的数据是数据闭环的首要任务;
④安全-紧要的场景是其中的“金矿”;
⑤仿真是解决自动驾驶安全的重要手段;
⑥自动标注加速整个开发流程。
谢谢大家

Contenu connexe

Tendances

Cruise AI under the Hood
Cruise AI under the HoodCruise AI under the Hood
Cruise AI under the HoodYu Huang
 
Autonomous Vehicle Research 2017
Autonomous Vehicle Research 2017Autonomous Vehicle Research 2017
Autonomous Vehicle Research 2017Brett Munster
 
Self-driving cars are here
Self-driving cars are hereSelf-driving cars are here
Self-driving cars are hereAndrew Ng
 
Autonomous vehicles
Autonomous vehiclesAutonomous vehicles
Autonomous vehiclesvishnum379
 
Vinod_Autonomous_car_ppts
Vinod_Autonomous_car_pptsVinod_Autonomous_car_ppts
Vinod_Autonomous_car_pptsvinumukkati
 
fusion of Camera and lidar for autonomous driving II
fusion of Camera and lidar for autonomous driving IIfusion of Camera and lidar for autonomous driving II
fusion of Camera and lidar for autonomous driving IIYu Huang
 
Self-Driving Cars With Convolutional Neural Networks (CNN.pptx
Self-Driving Cars With Convolutional Neural Networks (CNN.pptxSelf-Driving Cars With Convolutional Neural Networks (CNN.pptx
Self-Driving Cars With Convolutional Neural Networks (CNN.pptxssuserf79e761
 
Introduction to Graph Neural Networks: Basics and Applications - Katsuhiko Is...
Introduction to Graph Neural Networks: Basics and Applications - Katsuhiko Is...Introduction to Graph Neural Networks: Basics and Applications - Katsuhiko Is...
Introduction to Graph Neural Networks: Basics and Applications - Katsuhiko Is...Preferred Networks
 
Autonomous Driving
Autonomous DrivingAutonomous Driving
Autonomous DrivingUsman Hashmi
 
Winning Kaggle 101: Introduction to Stacking
Winning Kaggle 101: Introduction to StackingWinning Kaggle 101: Introduction to Stacking
Winning Kaggle 101: Introduction to StackingTed Xiao
 
Course on Optimal Transport
Course on Optimal TransportCourse on Optimal Transport
Course on Optimal TransportBruno Levy
 
Prediction,Planninng & Control at Baidu
Prediction,Planninng & Control at BaiduPrediction,Planninng & Control at Baidu
Prediction,Planninng & Control at BaiduYu Huang
 
Reinforcement learning, Q-Learning
Reinforcement learning, Q-LearningReinforcement learning, Q-Learning
Reinforcement learning, Q-LearningKuppusamy P
 
Azure AI platform - Automated ML workshop
Azure AI platform - Automated ML workshopAzure AI platform - Automated ML workshop
Azure AI platform - Automated ML workshopParashar Shah
 
Fault tolerant mechanisms in Big Data
Fault tolerant mechanisms in Big DataFault tolerant mechanisms in Big Data
Fault tolerant mechanisms in Big DataKaran Pardeshi
 
Machine Learning for Self-Driving Cars
Machine Learning for Self-Driving CarsMachine Learning for Self-Driving Cars
Machine Learning for Self-Driving CarsJan Wiegelmann
 
Face Detection techniques
Face Detection techniquesFace Detection techniques
Face Detection techniquesAbhineet Bhamra
 
[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review
[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review
[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper reviewtaeseon ryu
 

Tendances (20)

Cruise AI under the Hood
Cruise AI under the HoodCruise AI under the Hood
Cruise AI under the Hood
 
Autonomous Vehicle Research 2017
Autonomous Vehicle Research 2017Autonomous Vehicle Research 2017
Autonomous Vehicle Research 2017
 
Self-driving cars are here
Self-driving cars are hereSelf-driving cars are here
Self-driving cars are here
 
Autonomous vehicles
Autonomous vehiclesAutonomous vehicles
Autonomous vehicles
 
Autonomous car
Autonomous carAutonomous car
Autonomous car
 
Vinod_Autonomous_car_ppts
Vinod_Autonomous_car_pptsVinod_Autonomous_car_ppts
Vinod_Autonomous_car_ppts
 
fusion of Camera and lidar for autonomous driving II
fusion of Camera and lidar for autonomous driving IIfusion of Camera and lidar for autonomous driving II
fusion of Camera and lidar for autonomous driving II
 
Self-Driving Cars With Convolutional Neural Networks (CNN.pptx
Self-Driving Cars With Convolutional Neural Networks (CNN.pptxSelf-Driving Cars With Convolutional Neural Networks (CNN.pptx
Self-Driving Cars With Convolutional Neural Networks (CNN.pptx
 
Introduction to Graph Neural Networks: Basics and Applications - Katsuhiko Is...
Introduction to Graph Neural Networks: Basics and Applications - Katsuhiko Is...Introduction to Graph Neural Networks: Basics and Applications - Katsuhiko Is...
Introduction to Graph Neural Networks: Basics and Applications - Katsuhiko Is...
 
Autonomous Driving
Autonomous DrivingAutonomous Driving
Autonomous Driving
 
Winning Kaggle 101: Introduction to Stacking
Winning Kaggle 101: Introduction to StackingWinning Kaggle 101: Introduction to Stacking
Winning Kaggle 101: Introduction to Stacking
 
Course on Optimal Transport
Course on Optimal TransportCourse on Optimal Transport
Course on Optimal Transport
 
Prediction,Planninng & Control at Baidu
Prediction,Planninng & Control at BaiduPrediction,Planninng & Control at Baidu
Prediction,Planninng & Control at Baidu
 
Reinforcement learning, Q-Learning
Reinforcement learning, Q-LearningReinforcement learning, Q-Learning
Reinforcement learning, Q-Learning
 
Azure AI platform - Automated ML workshop
Azure AI platform - Automated ML workshopAzure AI platform - Automated ML workshop
Azure AI platform - Automated ML workshop
 
Fault tolerant mechanisms in Big Data
Fault tolerant mechanisms in Big DataFault tolerant mechanisms in Big Data
Fault tolerant mechanisms in Big Data
 
Machine Learning for Self-Driving Cars
Machine Learning for Self-Driving CarsMachine Learning for Self-Driving Cars
Machine Learning for Self-Driving Cars
 
Face Detection techniques
Face Detection techniquesFace Detection techniques
Face Detection techniques
 
Ebgan
EbganEbgan
Ebgan
 
[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review
[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review
[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review
 

Similaire à Data Closed Loop in Simulation Test of Autonomous Driving

AI-Car: Solution and Routes to accelerate Autonomous Driving's landing from r...
AI-Car: Solution and Routes to accelerate Autonomous Driving's landing from r...AI-Car: Solution and Routes to accelerate Autonomous Driving's landing from r...
AI-Car: Solution and Routes to accelerate Autonomous Driving's landing from r...KevinBai14
 
Spot basic info-2021.01.08
Spot basic info-2021.01.08Spot basic info-2021.01.08
Spot basic info-2021.01.08August Lin
 
海通证券金融云思考与实践(数据技术嘉年华2017)
海通证券金融云思考与实践(数据技术嘉年华2017)海通证券金融云思考与实践(数据技术嘉年华2017)
海通证券金融云思考与实践(数据技术嘉年华2017)Zhaoyang Wang
 
移动互联网服务端架构介绍
移动互联网服务端架构介绍移动互联网服务端架构介绍
移动互联网服务端架构介绍arganzheng
 
03_AI旅宿訂房系統-AI智慧訂房方案
03_AI旅宿訂房系統-AI智慧訂房方案03_AI旅宿訂房系統-AI智慧訂房方案
03_AI旅宿訂房系統-AI智慧訂房方案IttrainingIttraining
 
DDD x Architecture
DDD x ArchitectureDDD x Architecture
DDD x ArchitectureClark
 
海量計算的學習歷程分析與雲端資料庫管理系統Sqlmr appliance一體機開發計畫書 20140101
海量計算的學習歷程分析與雲端資料庫管理系統Sqlmr appliance一體機開發計畫書 20140101海量計算的學習歷程分析與雲端資料庫管理系統Sqlmr appliance一體機開發計畫書 20140101
海量計算的學習歷程分析與雲端資料庫管理系統Sqlmr appliance一體機開發計畫書 20140101Jackie Liu
 
冉有 支付宝钱包的研发挑战和最佳实践
冉有 支付宝钱包的研发挑战和最佳实践冉有 支付宝钱包的研发挑战和最佳实践
冉有 支付宝钱包的研发挑战和最佳实践Trinea Trinea
 
研華物聯網之無線監控技術發展現況_092111
研華物聯網之無線監控技術發展現況_092111研華物聯網之無線監控技術發展現況_092111
研華物聯網之無線監控技術發展現況_092111Eric Lo
 
Chap12 嵌入式系统开发模式
Chap12 嵌入式系统开发模式Chap12 嵌入式系统开发模式
Chap12 嵌入式系统开发模式Allen Galler
 
HR-032-資訊系進路圖
HR-032-資訊系進路圖HR-032-資訊系進路圖
HR-032-資訊系進路圖handbook
 
Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境
Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境
Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境drewz lin
 
微信201204
微信201204微信201204
微信201204drewz lin
 
微信之道201204
微信之道201204微信之道201204
微信之道201204shaomeng shi
 
未来网络技术发展探梦 - 开篇
未来网络技术发展探梦 - 开篇未来网络技术发展探梦 - 开篇
未来网络技术发展探梦 - 开篇Yao-Wei Ou
 
從雲端到邊緣 Azure IoT Edge 幫工廠設備長智慧
從雲端到邊緣 Azure IoT Edge 幫工廠設備長智慧從雲端到邊緣 Azure IoT Edge 幫工廠設備長智慧
從雲端到邊緣 Azure IoT Edge 幫工廠設備長智慧Poy Chang
 
2014 Hpocon 高驰涛 云智慧 - apm在高性能架构中的应用
2014 Hpocon 高驰涛   云智慧 - apm在高性能架构中的应用2014 Hpocon 高驰涛   云智慧 - apm在高性能架构中的应用
2014 Hpocon 高驰涛 云智慧 - apm在高性能架构中的应用Michael Zhang
 

Similaire à Data Closed Loop in Simulation Test of Autonomous Driving (20)

AI-Car: Solution and Routes to accelerate Autonomous Driving's landing from r...
AI-Car: Solution and Routes to accelerate Autonomous Driving's landing from r...AI-Car: Solution and Routes to accelerate Autonomous Driving's landing from r...
AI-Car: Solution and Routes to accelerate Autonomous Driving's landing from r...
 
Spot basic info-2021.01.08
Spot basic info-2021.01.08Spot basic info-2021.01.08
Spot basic info-2021.01.08
 
13 11-08
13 11-0813 11-08
13 11-08
 
2016 年春季嵌入式作業系統課程說明
2016 年春季嵌入式作業系統課程說明2016 年春季嵌入式作業系統課程說明
2016 年春季嵌入式作業系統課程說明
 
海通证券金融云思考与实践(数据技术嘉年华2017)
海通证券金融云思考与实践(数据技术嘉年华2017)海通证券金融云思考与实践(数据技术嘉年华2017)
海通证券金融云思考与实践(数据技术嘉年华2017)
 
移动互联网服务端架构介绍
移动互联网服务端架构介绍移动互联网服务端架构介绍
移动互联网服务端架构介绍
 
03_AI旅宿訂房系統-AI智慧訂房方案
03_AI旅宿訂房系統-AI智慧訂房方案03_AI旅宿訂房系統-AI智慧訂房方案
03_AI旅宿訂房系統-AI智慧訂房方案
 
DDD x Architecture
DDD x ArchitectureDDD x Architecture
DDD x Architecture
 
海量計算的學習歷程分析與雲端資料庫管理系統Sqlmr appliance一體機開發計畫書 20140101
海量計算的學習歷程分析與雲端資料庫管理系統Sqlmr appliance一體機開發計畫書 20140101海量計算的學習歷程分析與雲端資料庫管理系統Sqlmr appliance一體機開發計畫書 20140101
海量計算的學習歷程分析與雲端資料庫管理系統Sqlmr appliance一體機開發計畫書 20140101
 
冉有 支付宝钱包的研发挑战和最佳实践
冉有 支付宝钱包的研发挑战和最佳实践冉有 支付宝钱包的研发挑战和最佳实践
冉有 支付宝钱包的研发挑战和最佳实践
 
研華物聯網之無線監控技術發展現況_092111
研華物聯網之無線監控技術發展現況_092111研華物聯網之無線監控技術發展現況_092111
研華物聯網之無線監控技術發展現況_092111
 
Chap12 嵌入式系统开发模式
Chap12 嵌入式系统开发模式Chap12 嵌入式系统开发模式
Chap12 嵌入式系统开发模式
 
HR-032-資訊系進路圖
HR-032-資訊系進路圖HR-032-資訊系進路圖
HR-032-資訊系進路圖
 
Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境
Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境
Top100summit 腾讯-周健-服务化与体系化解决大量定制小项目开发困境
 
微信201204
微信201204微信201204
微信201204
 
微信之道201204
微信之道201204微信之道201204
微信之道201204
 
未来网络技术发展探梦 - 开篇
未来网络技术发展探梦 - 开篇未来网络技术发展探梦 - 开篇
未来网络技术发展探梦 - 开篇
 
從雲端到邊緣 Azure IoT Edge 幫工廠設備長智慧
從雲端到邊緣 Azure IoT Edge 幫工廠設備長智慧從雲端到邊緣 Azure IoT Edge 幫工廠設備長智慧
從雲端到邊緣 Azure IoT Edge 幫工廠設備長智慧
 
2014 Hpocon 高驰涛 云智慧 - apm在高性能架构中的应用
2014 Hpocon 高驰涛   云智慧 - apm在高性能架构中的应用2014 Hpocon 高驰涛   云智慧 - apm在高性能架构中的应用
2014 Hpocon 高驰涛 云智慧 - apm在高性能架构中的应用
 
About grow up
About grow upAbout grow up
About grow up
 

Plus de Yu Huang

BEV Joint Detection and Segmentation
BEV Joint Detection and SegmentationBEV Joint Detection and Segmentation
BEV Joint Detection and SegmentationYu Huang
 
BEV Object Detection and Prediction
BEV Object Detection and PredictionBEV Object Detection and Prediction
BEV Object Detection and PredictionYu Huang
 
Fisheye/Omnidirectional View in Autonomous Driving V
Fisheye/Omnidirectional View in Autonomous Driving VFisheye/Omnidirectional View in Autonomous Driving V
Fisheye/Omnidirectional View in Autonomous Driving VYu Huang
 
Fisheye/Omnidirectional View in Autonomous Driving IV
Fisheye/Omnidirectional View in Autonomous Driving IVFisheye/Omnidirectional View in Autonomous Driving IV
Fisheye/Omnidirectional View in Autonomous Driving IVYu Huang
 
LiDAR in the Adverse Weather: Dust, Snow, Rain and Fog (2)
LiDAR in the Adverse Weather: Dust, Snow, Rain and Fog (2)LiDAR in the Adverse Weather: Dust, Snow, Rain and Fog (2)
LiDAR in the Adverse Weather: Dust, Snow, Rain and Fog (2)Yu Huang
 
Scenario-Based Development & Testing for Autonomous Driving
Scenario-Based Development & Testing for Autonomous DrivingScenario-Based Development & Testing for Autonomous Driving
Scenario-Based Development & Testing for Autonomous DrivingYu Huang
 
Annotation tools for ADAS & Autonomous Driving
Annotation tools for ADAS & Autonomous DrivingAnnotation tools for ADAS & Autonomous Driving
Annotation tools for ADAS & Autonomous DrivingYu Huang
 
Simulation for autonomous driving at uber atg
Simulation for autonomous driving at uber atgSimulation for autonomous driving at uber atg
Simulation for autonomous driving at uber atgYu Huang
 
Multi sensor calibration by deep learning
Multi sensor calibration by deep learningMulti sensor calibration by deep learning
Multi sensor calibration by deep learningYu Huang
 
Prediction and planning for self driving at waymo
Prediction and planning for self driving at waymoPrediction and planning for self driving at waymo
Prediction and planning for self driving at waymoYu Huang
 
Jointly mapping, localization, perception, prediction and planning
Jointly mapping, localization, perception, prediction and planningJointly mapping, localization, perception, prediction and planning
Jointly mapping, localization, perception, prediction and planningYu Huang
 
Data pipeline and data lake for autonomous driving
Data pipeline and data lake for autonomous drivingData pipeline and data lake for autonomous driving
Data pipeline and data lake for autonomous drivingYu Huang
 
Open Source codes of trajectory prediction & behavior planning
Open Source codes of trajectory prediction & behavior planningOpen Source codes of trajectory prediction & behavior planning
Open Source codes of trajectory prediction & behavior planningYu Huang
 
Lidar in the adverse weather: dust, fog, snow and rain
Lidar in the adverse weather: dust, fog, snow and rainLidar in the adverse weather: dust, fog, snow and rain
Lidar in the adverse weather: dust, fog, snow and rainYu Huang
 
Autonomous Driving of L3/L4 Commercial trucks
Autonomous Driving of L3/L4 Commercial trucksAutonomous Driving of L3/L4 Commercial trucks
Autonomous Driving of L3/L4 Commercial trucksYu Huang
 
3-d interpretation from single 2-d image V
3-d interpretation from single 2-d image V3-d interpretation from single 2-d image V
3-d interpretation from single 2-d image VYu Huang
 
3-d interpretation from single 2-d image IV
3-d interpretation from single 2-d image IV3-d interpretation from single 2-d image IV
3-d interpretation from single 2-d image IVYu Huang
 
3-d interpretation from single 2-d image III
3-d interpretation from single 2-d image III3-d interpretation from single 2-d image III
3-d interpretation from single 2-d image IIIYu Huang
 
BEV Semantic Segmentation
BEV Semantic SegmentationBEV Semantic Segmentation
BEV Semantic SegmentationYu Huang
 
Unsupervised/Self-supervvised visual object tracking
Unsupervised/Self-supervvised visual object trackingUnsupervised/Self-supervvised visual object tracking
Unsupervised/Self-supervvised visual object trackingYu Huang
 

Plus de Yu Huang (20)

BEV Joint Detection and Segmentation
BEV Joint Detection and SegmentationBEV Joint Detection and Segmentation
BEV Joint Detection and Segmentation
 
BEV Object Detection and Prediction
BEV Object Detection and PredictionBEV Object Detection and Prediction
BEV Object Detection and Prediction
 
Fisheye/Omnidirectional View in Autonomous Driving V
Fisheye/Omnidirectional View in Autonomous Driving VFisheye/Omnidirectional View in Autonomous Driving V
Fisheye/Omnidirectional View in Autonomous Driving V
 
Fisheye/Omnidirectional View in Autonomous Driving IV
Fisheye/Omnidirectional View in Autonomous Driving IVFisheye/Omnidirectional View in Autonomous Driving IV
Fisheye/Omnidirectional View in Autonomous Driving IV
 
LiDAR in the Adverse Weather: Dust, Snow, Rain and Fog (2)
LiDAR in the Adverse Weather: Dust, Snow, Rain and Fog (2)LiDAR in the Adverse Weather: Dust, Snow, Rain and Fog (2)
LiDAR in the Adverse Weather: Dust, Snow, Rain and Fog (2)
 
Scenario-Based Development & Testing for Autonomous Driving
Scenario-Based Development & Testing for Autonomous DrivingScenario-Based Development & Testing for Autonomous Driving
Scenario-Based Development & Testing for Autonomous Driving
 
Annotation tools for ADAS & Autonomous Driving
Annotation tools for ADAS & Autonomous DrivingAnnotation tools for ADAS & Autonomous Driving
Annotation tools for ADAS & Autonomous Driving
 
Simulation for autonomous driving at uber atg
Simulation for autonomous driving at uber atgSimulation for autonomous driving at uber atg
Simulation for autonomous driving at uber atg
 
Multi sensor calibration by deep learning
Multi sensor calibration by deep learningMulti sensor calibration by deep learning
Multi sensor calibration by deep learning
 
Prediction and planning for self driving at waymo
Prediction and planning for self driving at waymoPrediction and planning for self driving at waymo
Prediction and planning for self driving at waymo
 
Jointly mapping, localization, perception, prediction and planning
Jointly mapping, localization, perception, prediction and planningJointly mapping, localization, perception, prediction and planning
Jointly mapping, localization, perception, prediction and planning
 
Data pipeline and data lake for autonomous driving
Data pipeline and data lake for autonomous drivingData pipeline and data lake for autonomous driving
Data pipeline and data lake for autonomous driving
 
Open Source codes of trajectory prediction & behavior planning
Open Source codes of trajectory prediction & behavior planningOpen Source codes of trajectory prediction & behavior planning
Open Source codes of trajectory prediction & behavior planning
 
Lidar in the adverse weather: dust, fog, snow and rain
Lidar in the adverse weather: dust, fog, snow and rainLidar in the adverse weather: dust, fog, snow and rain
Lidar in the adverse weather: dust, fog, snow and rain
 
Autonomous Driving of L3/L4 Commercial trucks
Autonomous Driving of L3/L4 Commercial trucksAutonomous Driving of L3/L4 Commercial trucks
Autonomous Driving of L3/L4 Commercial trucks
 
3-d interpretation from single 2-d image V
3-d interpretation from single 2-d image V3-d interpretation from single 2-d image V
3-d interpretation from single 2-d image V
 
3-d interpretation from single 2-d image IV
3-d interpretation from single 2-d image IV3-d interpretation from single 2-d image IV
3-d interpretation from single 2-d image IV
 
3-d interpretation from single 2-d image III
3-d interpretation from single 2-d image III3-d interpretation from single 2-d image III
3-d interpretation from single 2-d image III
 
BEV Semantic Segmentation
BEV Semantic SegmentationBEV Semantic Segmentation
BEV Semantic Segmentation
 
Unsupervised/Self-supervvised visual object tracking
Unsupervised/Self-supervvised visual object trackingUnsupervised/Self-supervvised visual object tracking
Unsupervised/Self-supervvised visual object tracking
 

Data Closed Loop in Simulation Test of Autonomous Driving